VG Pedigree Mapper, Simulation, and Variant Calling Evaluation Scripts

This repository contains scripts used to reproduce our work with the vg pedigree workflow in toil-vg to conduct the mapping, read simulation, and variant calling experiments.

Workflow Overview

The scripts should be run in the following order:

• Data preparation and graph construction workflow scripts
• Read simulation and simulated mapping evaluation scripts
• Mapping scripts, for assessing accuracy between VG Pedigree alignments and other competing mappers
• Variant calling scripts and evaluation, for assessing variant calling accuracy from alignments produced between VG Pedigree and other competing mappers

Output files

All output files by default will locate under directories placed within the path pointed to by the ${HOME} environment variable. Graph construction outputs will be located within ${HOME}/run_graph_construction. Simulated reads and mapping evaluations will be located within ${HOME}/run_sim_reads. Mapping output from BWAMEM alignments will be located within ${HOME}/run_bwamem_mapping. Mapping output from DRAGEN alignments will be located within ${HOME}/run_dragen_mapping. Mapping output from VG Giraffe alignments will be located within ${HOME}/run_giraffe_mapping. Mapping output from VG Pedigree alignments will be located within ${HOME}/run_giraffe_pedigree_mapping. Calling output and evaluation data from DeepTrio and Dragen variant callers will be located within ${HOME}/run_genotyping.

Replication Considerations

Though the top level workflows and scripts are automated, they have a number of system-level requirements. Access to a server that houses the Illumina Dragen Bio-IT Platform is required to run the DRAGEN-dependent scripts in scripts/wgs_mapping_experiments and scripts/wgs_calling_experiments. The toil-vg program used in scripts/graph_construction, scripts/wgs_mapping_simulation, and scripts/wgs_mapping_experiments requires a significant amount of memory and scratch disk space to run in single-machine mode that these scripts are configured to run on by default. A server with at least a 32 core CPU, 100GB of memory and at least 5TB of local disk space is recommended to run in single-machine mode. All software in these scripts rely on software containers in order to run, local installation of Docker is required to run them.